Some New Theories of the Greek Ka-Perfect. 
155 
origin only from such stems as possessed a gutturallv extended form by the 
side of the original one. Thus 7troo6dao for {fitTGo k-igo) which points plainly 
to a stem 7 traoK-, as an extension of the stem tct- ( cf. 7Ci-7tr-ao), gives the 
perfect TtaitTGau-a. This 7t£7trGOKa, being associated with 7ti-7tr-Go as its 
perfect naturally gave rise to the conception of -coxa as the perfect suffix, 
whence arose such forms as ol'xGoxa, possibly even kSijdoxa. 
This is intended to serve only as an illustration of the method to be fol¬ 
lowed in seeking the origin of the -na- perfect. Hartmann admits that the 
relation between tct- and TtaTCtooua could only give us -coxa as a perfect 
suffix (as seen in ox^xa, oi'xGoxa), not -xa. He is consequently led to seek 
further in search of some analogy which shall be more far-reaching in its 
effects. He thinks he finds this in the relations existing between fiddxGo 
and fdEfrjxa. He explains fiafirjxa as the perfect of fdddxco properly, not 
of fiaivao, and as arising by proportional analogy. Just as Xddxoo, which 
is a real guttural stem (being for *Xax- Oxgo, cf. s-Xax-ov, aorist), forms its 
perfect XeXrjxa, so after the same analogy fdddxai (though not really from 
a guttural stem) forms the perfect fdafh/xa. 
After fteftrjKa had once come into existence the analogy extended to 
other presents in - dxco, e. g., Srr/dxGD, ftXddxGD, fdi/UpGodxGo, etc., whence 
the perfects reSvr/xa, fiepfiXcoxa, fisfipGDxcc, then subsequently to other 
stems. , 
Such in brief is Hartmann’s theory, though his own statement of it is 
exceedingly unfortunate and obscure. It will be seen that it resembles 
Brugmann’s theory quite closely. I hold it, however, as much less proba¬ 
ble than Brugmann’s for three reasons. 
1. The analogy by which fiddxGo is supposed to form its perfect 
/Sefirjxa, viz., from the relation existing between XddxGo and XeXrjxa is 
not at all plausible. There is no similarity of meaning between the two 
words which would tend to associate them together in the mind, nor is 
XadKGD a common word or one which from the frequency of its use would 
be expected to exercise an influence upon other word-forms. 
2. There is no evidence that the form fiafipua is more recent than 
XiXrjxa, as is implied by the terms of Hartmann’s theory. Both forms are 
found in the oldest monuments of the Greek language, but nothing tends 
to show that XdXrjxa existed before /SefdrjKcx, so as to serve as the model 
for the formation of the latter. 
3. It is ncft likely that a formation taking its origin from a present with 
a short vowel, like fidduGo should operate immediately in affecting presents 
with a long vowel as Srr/duGo, fiXoadKGo, fiifSpaodKGo, etc., while at the same 
time failing to influence presents with a short vowel, such as fiodKGa, 
tpddKGD, etc. On the other hand we should expect these short vowel pres¬ 
ents to be first affected by the analogy, and the long vowel presents later, 
if at all. 
